Evaluating Surgical Results in Breast Cancer with Artificial Intelligence

Area of Science:

• Medical Artificial Intelligence
• Surgical Oncology
• Health Informatics

Background:

• Artificial intelligence (AI) is increasingly utilized in healthcare for applications such as surgical evaluation.
• Aesthetic symmetry is crucial for patient satisfaction and well-being following breast cancer surgery.
• AI offers potential for objective breast symmetry assessment, complementing human evaluation.

Purpose of the Study:

• To evaluate the performance of publicly available AI models in breast symmetry assessment.
• To compare these AI models against a custom-developed AI model (Pyolo8).
• To explore the emotional and ethical implications of AI-driven assessments for postoperative patients.

Main Methods:

• Sixty-eight patients undergoing breast reconstruction were assessed using publicly available AI models and the Pyolo8 model.
• Human observers evaluated all AI-generated results.
• Direct patient interaction with AI models was avoided due to misinterpretation risks.

Main Results:

• ChatGPT 4o and Pyolo8 demonstrated a statistically significant moderate to strong positive correlation with human observers for postoperative symmetry assessment.
• AI models showed potential for objective symmetry evaluation.
• Concerns regarding patient misinterpretation led to the exclusion of direct AI-patient interaction.

Conclusions:

• Both ChatGPT and Pyolo8 exhibit strong correlation with human assessments in breast symmetry evaluation.
• ChatGPT possesses superior communication capabilities compared to Pyolo8.
• AI systems currently lack the necessary empathy for direct patient interaction; human oversight is essential for safe and effective integration into patient care.